Stable Diffusion (family)

Stability AI's open-weight image generation diffusion model family — SD 1.5, SDXL, SD3, SD 3.5 — the foundation of the open-source AI art ecosystem.

Stable Diffusion is the open-weight image generation model family from Stability AI (with significant external collaboration on early versions, including Runway and CompVis). The line: Stable Diffusion 1.4/1.5 (2022, the breakthrough that made AI image gen accessible to everyone), SDXL (2023, much higher quality), SD3 (2024, multimodal diffusion transformer architecture), SD 3.5. It matters because Stable Diffusion is to image generation what Llama is to LLMs — the open foundation that an entire ecosystem is built on. Hundreds of fine-tuned variants on Civitai and Hugging Face, the entire ControlNet ecosystem, AnimateDiff for video, and most non-commercial AI art tools trace back to SD checkpoints. ComfyUI and Automatic1111 web UIs exist because of SD's open weights. SD 1.5 in particular has been widely customized — anime models, photorealistic models, character LoRAs, style LoRAs. The model is small enough (~4 GB) to run on a consumer GPU, generating images in seconds. Newer competitors in the closed-source frontier image generation space — DALL-E 3, Midjourney, Imagen, Flux Pro — generally produce higher absolute quality. But for customization, fine-tuning, controlled generation, and offline use, the SD family remains foundational. Related: diffusion model, image generation, ControlNet, LoRA, Flux.